Announcement
Collapse
No announcement yet.
Catalyst 14.4 Has Advantages Over Linux 3.15 + Mesa 10.3 Git
Collapse
X
-
Originally posted by dungeon View PostI overall considered radeon driver at 40-80% GL performance level of Windows Catalyst. And we are at the same level for more then 10 (fuckin' years baby yes ) years, so nothing changed for that matter in more then 10 years . I use r200 driver ~10 years (and compared it from the dri1 times when it perform the best that is around 2006. just before AMD cames with their opensource strategy), and now i see radeonsi is on the same level when compared with Windows Catalyst, so i can clearly say - absolutly nothing changed there with GL performance wise .
When i say that i always excluded bugs in radeon (there are many bugs out there), i just considered that N/A .
10 years ago, hd4830 with no DPM did 1 fps software rendering with radeon; and 300 fps in catalyst, with 40fps at pixel_vomit map with bloom+flares+hires textures.
Today hd830 with radeon has automatic DPM and it does 280 fps and 40 fps at pixel_vomit map with bloom+flares+hires textures (since kernel 3.15+Mesa 10.2.2).
Alone the fact how open radeon destroyed closed source macos, which has similar marketshare to linux, shows a LOT.
Your R100 card is over 10 years old, what improvement do you expect? Pick more modern card and do a time travel. This is considering Linux.
With windumbs, there is no point in time travel because it has HUGE marketshare. So referencing windumbs has only a point as a maximum achievable performance and not as a usable alternatives. Because in that case, thats completely wrong forum.
Comment
-
Originally posted by verde View PostI strongly believe after all those years that opensource AMD drivers will NEVER be on par with the closedsource when we talk about Gaming.
Even when they perform better, is because tesselation and shaders are missing.
And then because of the quality of the AMD closed-source drivers, I chose nVidia GPU again. Their drivers are in excellent shape and their legacy support very good.
I was surprised how well the modern intel gpus run too with mesa
Comment
-
-
Originally posted by brosis View PostOpenarena:
10 years ago, hd4830 with no DPM did 1 fps software rendering with radeon; and 300 fps in catalyst, with 40fps at pixel_vomit map with bloom+flares+hires textures.
Today hd830 with radeon has automatic DPM and it does 280 fps and 40 fps at pixel_vomit map with bloom+flares+hires textures (since kernel 3.15+Mesa 10.2.2).
Alone the fact how open radeon destroyed closed source macos, which has similar marketshare to linux, shows a LOT.
Your R100 card is over 10 years old, what improvement do you expect? Pick more modern card and do a time travel. This is considering Linux.
Maybe you don't understand what i am saying... In year 2004. if you ware buying R200 card to use radeon driver, and now in year 2014. if you buy current hardware, you will have aproximately same raw performance if you compare Catalyst for Windows against Mesa radeon drivers .
With windumbs, there is no point in time travel because it has HUGE marketshare. So referencing windumbs has only a point as a maximum achievable performance and not as a usable alternatives. Because in that case, thats completely wrong forum.Last edited by dungeon; 19 July 2014, 12:11 AM.
Comment
-
Originally posted by dungeon View Post10 years ago, we needed to wait about 5 years more for hd4830 to be even released, let alone being (good) supported by Mesa drivers .
Maybe you don't understand what i am saying... In year 2004. if you ware buying R200 card to use radeon driver, and now in year 2014. if you buy current hardware, you will have aproximately same raw performance if you compare Catalyst for Windows against Mesa radeon drivers .
They've also been playing catch-up with OpenGL support, based upon what I can see I'm guessing it's going to be at least another year or two before we get to OpenGL 4.4. Once OpenGL is up to date, and OpenCL has been implemented, and whatever other important techs come up do I expect to see the radeon driver being optimized without a significant increase in manpower I personally find it ridiculous that you expect catalyst level performance before the OpenGL support is complete. As for me since the dynamic power management work hit I find that my experience is much better than either Catalyst or Nvidia provided even if raw performance is still lacking in comparison and will continue buying AMD hardware for the foreseeable future.
Comment
Comment